🇺🇸Kaitlyn
Wang

The star catalog had been sitting in a NASA database for years — 200,000 light-curves, the faint brightness flickers of distant suns, each one potentially concealing a world. Searching it for planets took 4.5 years using conventional methods. Then a seventeen-year-old from San José decided that was unacceptable. In May 2023, Kaitlyn Wang walked into the Regeneron International Science and Engineering Fair in Dallas and walked out with $75,000 and the top prize in the world — not because she was lucky, but because she had written an algorithm that did in fourteen days what previously took forty-five centuries of human patience.

Wang attends The Harker School in San José, one of Silicon Valley's most academically rigorous independent schools, but her obsession with exoplanets began long before any classroom assignment. The exoplanet field — the study of worlds orbiting stars other than our Sun — has exploded since NASA's Kepler and TESS missions began generating vast archives of stellar light data. The challenge is signal extraction: buried in the noise of starlight is the almost imperceptible dimming that occurs when a planet passes in front of its host star, an event called a transit. Finding ultra-short-period planets, those completing a full orbit in less than a day, is particularly grueling because their transits are brief and their signals nearly invisible.

Wang's solution — ExoScout — combined two powerful computational ideas. At its core is a phase-folding algorithm accelerated by a Graphics Processing Unit (GPU), a chip originally designed for video games but repurposed here to crunch stellar data in parallel. Layered on top is a Convolutional Neural Network (CNN), a type of artificial intelligence trained to recognize the characteristic shape of a planetary transit the way a radiologist learns to read an X-ray. The CNN sifts through the GPU's output and flags genuine planetary candidates with 97 percent accuracy, discarding the false positives that plague conventional searches.

The results were not theoretical. Wang ran ExoScout against the Kepler Input Catalog — that 200,000-star database — and found three planets that had never been catalogued. One of them, Kepler-1598d, is the smallest ultra-short-period planet ever discovered. It completes an orbit in roughly eighteen hours. Its year is shorter than most people's workweek. The star it circles is older than the Sun, which raises extraordinary questions about the planet's history, its surface conditions, and whether any such world could ever have harbored the conditions for life.

The judges at ISEF — scientists and engineers from industry and academia — awarded Wang the George D. Yancopoulos Innovator Award, named for Regeneron's co-founder and chief scientific officer. The prize of $75,000 is the single largest individual award at the fair. Later that year, she was named a Davidson Fellow, one of the most prestigious recognitions for young scholars in the United States, in a cohort reserved for research of genuine national significance.

Wang's work belongs to a tradition of citizen astronomy that has quietly reshaped the field — the understanding that the universe's secrets are not locked away in elite observatories but encoded in publicly accessible databases, waiting for someone clever enough to ask the right computational question. She has made that access democratically real. ExoScout runs on a laptop-grade GPU. A well-equipped school with a committed student could run it. In a field where the cost of discovery has historically scaled with institutional budgets, that is a genuinely radical idea.

Beyond the technical achievement, Wang represents something important about how science is changing. She was in eleventh grade when she did this work. She had not yet taken a university course. She taught herself the convolutional neural network architecture by reading papers and documentation, built the GPU pipeline on consumer hardware, and validated her results against the existing Kepler catalog before submitting to ISEF. The precision, independence, and ambition of the project would be impressive in a PhD student. In a high schooler, it is extraordinary.